Public transit agencies and private firms have decided that a major problem facing cities is the “first mile/last mile” (FMLM) problem. The FMLM problem is drawn originally from telecommunications, then supply chain management (goods movement). For telecommunications FMLM is the final leg (or first leg) to the consumer. With physical infrastructure it is expensive to match high capacity hubs to individual units. In the 1970s and 1980s, as cable TV was being deployed across the US, cable companies had to individually wire each and every household. This was a tremendous but necessary cost, and the cable companies were able to amortize the expenses over many years. Rarely did an individual household pay the full cost of running cable, instead they just paid a small installation fee and their monthly subscription.

FMLM then was used by logistics companies (FedEx, UPS, etc.) to describe their end point deliveries from centralized warehouses. These trips are complex chains to optimize, and ultimately the goal is to lower the cost as much as possible to lower the overall cost of shipping a parcel from Point A to any other Point. For both telecommunications and logistics, FMLM has large literatures of public research plus scads of private research.

In recent years FMLM has been used to also describe passenger travel in the context of getting to/from bus and rail stops. While there are some similarities between telecommunications, goods movement and passenger travel, FMLM for transit isn’t well understood. A Google Scholar search suggests that there are few studies overall, and even fewer peer reviewed studies. Of what is out there, most have considered bicycles (or Segways!) as the preferred FMLM solution.

The dearth of FMLM research isn’t problematic on its own. Lots of things that should be studied haven’t been studied much, and lots of things that we really don’t need to know more about keep getting poked and prodded as though something new will fall out. But for FMLM for transit, not knowing anything about what makes it successful is a bit of a challenge, especially since there are lots of private firms now competing for scarce public dollars to subsidize favored technologies and modes. We should support the spirit of innovation at the heart of these efforts to boost transit ridership. But we should also insist that solid social science research comes out of the many pilots underway.

What I have seen so far is that there are many transit agencies and cities starting to look to Uber, Lyft, Juno, etc. to provide short trips to and from transit stops (mostly rail from what I can tell). These taxi trips will be subsidized and hopefully enhance existing transit service. So far, so good. Let’s all agree more transit ridership is better than less transit ridership.

Beyond these pilots, however, what do we know? Here are some issues at hand:

• Elasticity of demand with regard to fares: Transit ridership is sensitive to fares. As fares go up, demand for transit goes down. (I’ll leave aside cross-elasticities and such for now, but they are real) For many reasons I am not convinced we can isolate a reliable point estimate of elasticity, but -0.3 is typical. A 1% increase in fares reduces demand by 0.3%. Will people pay additional taxi fares, even if subsidized, to take transit they are paying a separate fare for?
• Disutility of transfers: We know that transit riders prefer to avoid transfers when possible, but many are willing for one transfer. Transfers have to be easy in time and effort. With long headways (up to 20 minutes) it is not clear that transfers between FMLM can be reliable enough to encourage use. Getting to the station is one problem. Waiting at the station is another.
• Opportunity cost: How much are transit agencies or cities willing to pay for what will be marginal improvements in ridership and are there better uses of those dollars? I suspect taxis to transit will have a marginal effect on ridership because even if you have a handful of drivers zooming around with passengers these will be one way trips to/from the station due to peaking. So maybe a driver can carry 4-5 passengers an hour if each round trip takes 12 minutes or so, and that’s assuming that people are ready when the taxi is. A van going from a train stop to a large employer, such as what happens in Silicon Valley, will have a somewhat higher capacity. These are uncommon cases. Bike share is also unlikely to have a large effect because of the one-way travel and imbalance problems. Even if I’m off by an order of magnitude that’s not very many people per hour getting on transit at any given stop. Some of the pilot projects have a one-year budget of about $200,000 (which doesn’t suggest a lot of trips. Assuming no administration costs and $5 subsidy per trip that’s 40,000 trips, or 3,333 per month or about 110 per day. So round trip it’s about 55 people.) That’s not much relative to a total transit agency operating budget, but it’s enough to improve headways or do something else that may also have a positive effect on ridership. In most places that’s close to enough to hire a few drivers and buy a bus or two (over a few years). Would ridership improve more by reducing headways or giving cheap taxi rides? That’s a question to answer.

Overall, FMLM is an idea that sounds just plausible enough to make sense intuitively. Yet we have to evaluate these projects in terms of bang for the buck. These pilots are worth pursuing, but they may or may not be cost effective in the end.

You can’t swing a dead cat without hitting another story about driverless cars, shared taxis, new mobility or other tech oriented developments in passenger travel. Invariably, these stories treat the technological advance as the Big Change for travel, where being able to summon a cab on your mobile phone changes everything we know about the world. However, looking at the future of transportation through just a technological lens misses the biggest economic shift underway—the shift from high average cost/low marginal cost travel to low average cost/high marginal cost travel. While this shift is enabled by technology, it is not just technology.

Eliminating drivers will not automatically reduce transportation costs to something approaching zero (I will explain this in a different post, but the short explanation is that drivers do far more than just drive.).  Cars will still cost money to build, maintain and operate. Recently I saw a quote from Emily Castor of Lyft where she said even if autonomous vehicles cost $500,000 each the cost of a Lyft ride will remain the same as it is today. This claim suggests that even companies in the middle of “disrupting” transportation don’t really know where the field is going or what the relationships between capital and operating costs are. In short, capital costs cannot be separated from operating costs.

Consider the current cost of driving in the US. Once you buy and insure a car, driving is extremely cheap for most trips. Most trips begin and end with free parking, and to drive a few miles requires fuel costs measured in nickels and dimes, not dollars. (A car that gets 20 miles per gallon will use $0.50 in fuel for a five mile trip if gas is $2 per gallon, for instance.) Automobility in the US features high average costs (some fraction of the fixed costs of buying and insuring plus the cost of travel) and very low marginal costs (the cost of an additional trip).

Now consider the cost of taxis (or transit). You don’t pay any fixed costs, but the cost of an individual trip is high. A taxi trip of five miles might cost $10 or more, where the same marginal cost for driving was $0.50. A transit trip will cost a few dollars as well, plus additional travel time. Considering these differences in marginal costs, I may drive for a bag of potato chips if I’m in the mood, but I’m not going to pay an Uber or taxi to take me to the Circle K and back. Potato chips aren’t that valuable.

This shift in costs will affect the travel we undertake, and many firms will aim to get the costs of travel as low as possible. In short, as the cost of travel increases per trip, we will travel less.

As an example, when I lived in New York City, I would sometimes use Amazon Prime Now for two hour deliveries since they charged $6 to deliver while my subway fares would have been $5.50. Now that I live in Phoenix, I either ride my bike or make a short drive with free parking that saves me most of the delivery fee. In NYC I transferred my trip to a delivery person. Hopefully that person will use the transport systems more efficiently than a bunch of individual travelers out for sacks of coffee beans.

The shift to higher trip costs that reflect fixed and operating costs will be the factor that affects how and how much we travel in the future. If trip costs are high, then we will travel less and demand for effectively zero cost travel—think walking and biking—will increase. If trip costs remain low, then we will travel more by car.

How low can firms get the costs of travel? Probably not as low as they think. There are cost they don’t control: including paying for infrastructure through fuel taxes, road pricing, parking charges, congestion pricing, etc. Plus someone has to buy and maintain all those vehicles that will get hired, and those costs won’t get lost in the haze of public subsidy (hopefully). Somebody will want a return on their capital investment—just like a real company!

(This is a guest post by David King)

Many planners argue for greater integration of transportation and land use policy. In the US, many argue that separated policy realms contributed to the system of automobility and low density cities that exist, in part because transit investment isn’t able to capitalize real estate benefits. This is different than the historical connections between streetcar companies and real estate development in nascent suburbs. In recent years transport planning, and transit planning in particular, has attempted to reshape the urban landscape through new infrastructure investment. A common refrain is that real estate developers love the “permanence of rail” when it comes to making location decisions about where to build. Another way strong connections between transportation and land use are promoted are through value capture mechanisms for road or transit projects. Overall, the contemporary debates about infrastructure tend to revolve around “how much” change to the built environment will be caused by new investment. This assumes a strong and measurable relationship between transportation and land use. Better questions may be: 1) whether the relationship is strong enough to affect change at all, and 2) are the connections between transportation and land use strengthening or weakening?

A weakened connection between transportation and land use is not a new idea. Gen Giuliano, for one, wrote about it in the 1990s. Marlon Boarnet argued more recently that changes in transportation will be a much bigger deal than changes in density moving forward. In short the argument is that ubiquitous automobility has reduced the onetime spatial advantages due to proximity to a limited number of transit options. At present there are good reasons to improve our understanding of how transportation and land use interact. Broadly speaking, there are two current approaches to transportation. First, there is massive public investment in a limited number of infrastructure projects. Such projects—light rail, commuter rail, streetcars, road expansion, etc.—represent a form of spatial planning where favored locations are gifted desirable infrastructure as amenities, with some expected transportation improvement that is mostly measured in travel time. Second, there are countless private firms working on selling personalized mobility. Taxi companies, tech-taxi companies, private bus companies, commuter shuttle services, vanpools, shared vehicles and others are all competing for passenger travel business. These companies are reacting to existing land use development and taking advantage of existing infrastructure to improve access by personalizing mass transit. Plus existing and new automakers are adapting their cars and business models to current and future consumer preferences and trends, and of course self-driving cars are coming as well.

The public model of large investment in a limited space stands in contrast to the private model of many alternatives working to customize transit to maximize accessibility across regions. The public clearly believes that the transportation and land use connections are strengthening, and the private firms are indifferent or they see a weak connection moving forward, which is why “real time data” is so important to everyone. A weaker transportation and land use connection suggests that spatial planning through infrastructure (see here for an Australian report on this issue, though this applies globally) will be ineffective, which has implications for local land use policy and transportation finance.

In earlier posts I highlighted the breadth of complementary transit systems that exist along with primary ones in New York City, and in another I proposed that transport project evaluation should account for pricing and zoning policies in place before new investment is made. Here what I argue is that the complementary aspects to any primary transport system are nearly always local concerns, and too little is known about how complementary local policies can make or break successful primary systems.*

One way to think about urban transport networks is that there are primary systems complemented by secondary systems. The primary systems are the highest capacity networks, spatially fixed and designed to broadly serve the population of the region. Freeway networks, rail transit and bus rapid transit are examples of primary systems in this way. Secondary systems are those that serve smaller, sometime niche, populations and act as complements to the primary systems. These secondary systems are critical to the overall success of the transport networks as primary systems are very good at serving predictable travel such as rush hours but less adept at serving lots of different types of trips to lots of different places.

One reason automobility is so pervasive is that the primary systems (freeways and other large roads) are nicely complemented with hosts of local policies that make driving really easy. Parking requirements, street design, signal timing and other aspects all come together to create a seamless driving experience in most US cities. No other travel mode has such an advantage and transit, taxis, cycling and walking are too often left to cobble together whatever kind of system they can. Rarely are these “alternative” modes granted the luxury of integrated primary and secondary systems, and even more rare are integrated transportation providers that manage all primary and secondary transportation. Transport for London is an example of an integrated approach.

Consider the U.S. investment made in transit over the past few decades. Transit now accounts for close to 20 percent of total Highway Trust Fund spending, yet the share of travelers using transit has barely moved nationally. This is not an indictment of transit as a worthy mode, just suggestive that we can spend that money more wisely. New York City is a true success story for transit ridership yet has received very little new investment for system expansion relative to maintenance costs (mostly through state of good repair programs) and broadly supportive complementary systems of taxis, liveries, jitneys and 24-hour service. Parking is also very expensive in New York and often not required for new construction and adaptive reuse. Road tolls—which account for about 25 percent of all road tolls collected in the country—have increased dramatically over the past few years. Over the past decade transit fares have been followed by increased rail ridership, though fares have largely kept pace with inflation since 2003. This does mean that New York’s relatively high farebox recovery rates —for the U.S.— have remained an important source of operating revenues. Taken together New York is doing lots of different things that support robust transit services that have little to do with the built environment and density, though these obviously help, but lots to do with complements to the primary systems.

Other cities certainly do a lot to support transit investment, but none support alternatives to the auto to the same degree that autos benefit from complementary policies. One challenge for cities is that the available complementary systems are not very well understood. As one example, current policy debates about taxi services and app-based ridesharing companies often minimize the complementary role of taxi services for transit-oriented cities. Even car sharing studies have largely viewed car sharing as a replacement for auto ownership (e.g. “how many cars does a shared car replace”) rather than an extension of transit policy (“how can shared cars extend access for transit users?”**). Obviously there are nuanced analyses of these complementary systems and the researchers involved know this. The point is that many systems that are sometimes considered substitutes (car sharing or taxi services) are more accurately considered complements, and necessary but not sufficient systems to support primary transit and road networks.

Another challenge for complementary systems is that autos are sufficiently utilitarian that once a city is designed for cars then all trips can be made with one vehicle. Auto drivers are unimodal, whereas transit riders, walkers and cyclists tend to use many more modes through the courses of their lives. This means that complementary systems for non-auto uses require many different types of investment, many of which will have small obvious payoffs. Less obvious payoffs may be large, however, which is a large reason we should study complementary modes more. For instance, based on work by Dan Hara in San Francisco we know that people are much more likely to take transit to work if they know they can get home by taxi if they have to work late. This holds even though most never use the taxi option. In New York it is common for office workers to get a guaranteed taxi ride home if they work past a certain hour (9pm in many cases). This program leverages the availability of complementary travel modes to reinforce the primary transit modes.

By focusing on complementary secondary transport systems cities can lay the groundwork that supports investment in primary systems. Investment in primary systems alone will not produce modal shifts. What if Interstate freeways had been built without minimum parking requirements as part of the zoning code, or a hierarchy of road networks designed to funnel drivers onto ever-faster roads? US cities would look and function very differently. The policy shifts toward automobility was comprehensive and included federal, state and local policies all working together to complement each other. Such voluntary coordination simply does not exist for any other mode.

*A search for “complementary transportation” returns pages of results that misspell “complimentary”.

**Many car share studies and advocacy pieces do mention access to cars for those who don’t have one as a benefit, but most research examines how car share can reduce the need to drive by car owners.

Of the many shifts in travel trends, one that gets a lot of attention is fewer miles traveled by auto. While some explanations for the reduction focus on factors of the built environment or demographic changes, the changing structure of daily travel trips receives less attention. Overall people of all types are making fewer trips per day, though the modal mix of daily journeys is changing. Commute trips and shopping trips have declined more than others. These are trips that are more easily substituted with not making a trip at all, either by working remotely or increased retail deliveries.

Fewer trips per traveler affects the overall cost of travel but potentially in more nuanced ways than expected. If a commuter starts working remotely once or twice a week, the math supporting their decision to buy and use a monthly pass for parking or transit changes. In particular, as people travel less per month the discounts from monthly passes for parking or transit decline. As “unlimited” passes create incentives to travel more as the cost per trip declines with use, as travel decreases the incentives to travel even less kick in. At some point, and this depends on many factors, the change in travel cost per trip paid via monthly pass outpaces the change in travel.

Say a monthly pass for unlimited transit is $100, and the traveler commutes 20 days per month (I use 20 as an even number; most full time workers commute 21 or 22 days). This means the traveler pays an average cost of $5 per day to commute (which is a not an atypical round trip transit fare), but all trips, including any additional trips above 20 have a marginal cost of $0. As long as the daily cost is more than $5 it is worth a monthly pass. But if the number of commutes is reduced by four as the traveler now works at home one day per week, the daily cost of the pass is $6.25. As this is likely higher than local round trip transit fare the commuter no longer buys the monthly pass. In addition, without the pass, the cost of any trip beyond the commute trips increased from a marginal cost of $0 to whatever a round trip transit fare is. Of course, if the numbers of marginal trips are known ahead of time perhaps the traveler will continue to buy the monthly pass.

Otherwise, the new cost structure of trips may accelerate switches to substitutes, mostly to not traveling at all. If each commute trip has a marginal cost then commuters can minimize this cost by taking fewer commute trips, so we should expect even more working at home. If each non-commute trip has a high marginal cost then people will also substitute away from these trips. Taken together, fewer trips may lead to fewer monthly passes, which raises the marginal cost of travel and creates a cycle where higher costs per trip further reduces the number of trips taken. You can draw a similar example for driving.

In limited discussions with people who work for large transit agencies they have indicated that ridership increases are mostly through individual fares (including bonuses) rather than monthly unlimited passes. I have not seen these data published, but if true it suggests that shifts in the frequency of travel may have effects on fare policy for transit. Another effect may be that transit agencies may get more money per trip if the unlimited trips are eliminated, but this may be offset by fewer trips per person. There also may be rapid shifts toward substitutes as increased travel costs per trip are not a linear function.

Once travelers stop buying monthly passes for transit or parking (e.g. buying in bulk) they may eagerly seek out alternatives because the marginal trip suddenly becomes much more expensive. For drivers this substitute may be mass transit, but it also may be fewer trips by all modes as they look to telecommute and shop online. It is also possible that the increment between marginal auto trip cost and marginal transit trip cost may not be that great for local trips after monthly passes no longer make sense.

All of this is speculative, but the takeaway is that as travel behaviors change, the costs of individual trips change, and we don’t have that clear an idea how this will all play out. We should keep an eye on sales of monthly passes as people travel less. The financial models for transit, parking and even toll roads may not be as predictable as we think.

US cities and regions have spent billions of dollars investing in new rail and bus transit with the hopes of luring drivers from their cars, reducing the environmental damage from personal transportation, minimizing congestion and encouraging economic development. Despite the enormous investment in transit over the past few decades, most cities have seen very little overall transit ridership gains, and the mode share for transit commuting remains about where it was in 1980. A puzzle associated with less than hoped for effects from increased supply of high quality transit is what, if any, complementary policies are cities implementing to raise the price of driving, increase transit supportive land uses, and provide many complementary modes of travel to automobility.

The way we evaluate transit (and transport projects overall) may be part of the problem. Recent work by Kate Lowe and Sandra Rosenbloom highlights that the single most important factor for federal funding of transit projects through New Starts is whether the local funding match is adequate. The next most important factor for funding is the project justification based on a range of predicted outcomes. As Lowe and Rosenbloom write:

“The new regulations permit or encourage applicants to claim a wider range of benefits as part of project justification while reducing the project costs against which benefits are evaluated. These changes are the source of great controversy. Some groups argue that they are long overdue because major transit projects have greater benefits than the FTA has recognized in the past, such as economic development and lowered environmental pollution. Other groups argue that that many of the claimed benefits are prime examples of the “optimism bias” to which the supporters of rail transit projects fall prey, overestimating the effect of transit improvements on ridership (Federal Transit Administration 2008) and attributing to them such benefits as decreased congestion and more concentrated and diverse land use patterns.”

There is nothing wrong with local matching funds or economic development. These are often important planning and governance goals. Yet the primacy of local revenues independent of transport benefits leads to new broadly applied taxes—most often sales taxes—that are only loosely connected to mobility and accessibility improvements. Economic development is also a worthy cause and a major part of local planning. However, transport projects for economic development should be evaluated against other policies and investments that promote economic development. With mature transport networks new economic development effects from an improvement are generally weak, though there may be stronger redistributive effects. Of course, redistribution creates winners and losers.

In many cases benefit/cost analyses are used to evaluate projects. These analyses are worthwhile but subject to political shenanigans and questionable assumptions. More troubling, the actual benefits and the costs of projects are rarely, if ever, evaluated ex post so there is little incentive to not exaggerate benefits and minimize expected costs.

So project evaluation is not perfect, and in many cases not that good, though we do seem to be getting better. However, the current system encourages pursuit of easy money rather than effective policy. There has to be a better way to allocate investment, and this is what I propose. Transport investment should be made on the basis of improved accessibility, congestion reduction, reduced negative externalities and economic efficiency. Rather than base federal spending on whether local governments can raise money or potentially realize benefits at some point in the future, federal spending (and other matches) should be based on what policies local governments have implemented to improve their transportation system.

Criteria used for evaluation for all new transport infrastructure investments should include whether cities and regions have done the following:

• Enacted congestion pricing. If a city, region or state charges directly for road use, with prices varying by location and time of day, this establishes that congestion is a problem and that the politically difficult work has been done.
• Relaxed zoning codes to allow greater densities and mixed uses. Again, these are politically difficult but the benefits are potentially large even without new transit or other transport investment because of improved walkability and more efficiently supplied public services.
• Eliminated parking requirements. At the very least, all parking requirements in designated transit corridors should be eliminated.
• Charged market prices for curb parking. No free parking. In fact, local policy should reflect the true cost of driving.
• Installed dedicated bus lanes. If a city or region will not, as a bare minimum, reallocate some road space to transit then why should they be rewarded with new spending? This is not to say all bus lanes must have their own right of way. But high capacity corridors should, and if high capacity corridors cannot be identified then that is evidence that no new spending for expansion is necessary.
• Liberalized the taxi market. This one is a bit tricky as we don’t know what an ideal taxi market should look like, but in too many cities the regulators are captured by the regulated.
• Met minimum farebox recovery standards. Farebox recovery in the United States is awful. Low farebox recovery makes planning new lines difficult as system expansion makes existing operating budgets worse. This is not to say that transit should have 100% farebox recovery, but a minimum requirement of at least 50% will help ensure that projects will be considered based on their actual demand rather than some future, unknown demand.

There are other potential criteria, too, but the point is that investment subsidy should be based on revealed actions and a willingness to take on the hard political work of improving our cities. Even if a city did all of the things listed above and didn’t get any money they would be better off because they did all of the things that are proven to improve transportation systems. But if cities do the things listed above, then it doesn’t really matter who pays for new investment as the new investment will very likely be worthwhile because the complementary policies are in place. If cities do the hard political work they should be rewarded. If all they do is raise taxes based on specious claims, they should be held accountable. We currently have this backwards.